DISGENET is provided by
Medbioinformatics
Solutions S.L.
Rambla de Cataluña 14, 7º 1ª
08007 Barcelona
(Spain)
Contact us:
info@disgenet.com
Target selection remains a key determinant of success in pharmaceutical R&D, as efficacy and safety issues continue to drive late-stage failures. In 2025, the FDA approved 46 drugs (Mullard 2026) (Table 1), including 38 protein-targeted therapies and three mRNA-based treatments, offering a real-world snapshot of which targets actually make it to the finish line. By integrating data on disease associations from DISGENET (Piñero et al. 2026), gene expression, and historical target status, we explored what these approvals can teach us about choosing the right targets.
All drug targets are represented in DISGENET, underscoring their established relevance to human disease. Most are associated with hundreds of disease phenotypes. All are referenced in hundreds, often thousands, of publications in the disease genomics literature, reflecting decades of accumulated biological and clinical knowledge. Consistent with this depth of evidence, many 2025 FDA approvals focus on long-studied targets such as EGFR, ERBB2, and ESR1, which exhibit publication histories dating back to the 1970s and broad links to diverse disease areas.
Figure 1. Target discovery timeline and disease associations. Each point represents a drug target approved in 2025. The x-axis shows when the target was first linked to disease in the literature (year of first PMID), the y-axis shows the number of diseases associated with the target, and point size reflects the total number of publications (PMIDs).
DISGENET also shows that most of the approved targets appear in clinical trial contexts. Examples include patient stratification (e.g., JAK2, MET) and monitoring pharmacodynamic response (e.g., APOC3, PCSK9, BTK), reflecting their translational relevance throughout drug development.
Figure 2. Growth of clinical trial activity by target over time. Cumulative number of clinical trials for selected targets in DISGENET. The lines show how clinical investigation has expanded for key targets approved in 2025.
To better understand why many targets are associated with a large number of diseases, we examined disease-class enrichment for each gene. Instead of counting diseases, we looked at whether a target’s associations repeatedly fall within the same therapeutic areas compared with what is typically seen across all genes in DISGENET. This shows that these broad associations are often concentrated within a small number of related disease classes that, in most cases, include the therapeutic indication of the target.
Representative examples illustrate this pattern (Figure 3). PCSK9 is linked to many diseases, but these are mostly related to hypercholesterolemia and related cardiovascular or metabolic disorders. Similarly, IL5 is enriched in immune, hematologic, and respiratory disease classes, consistent with its role as the target of depemokimab (Exdensur) for severe eosinophilic asthma. Together, these examples show that many targets are not broadly associated with unrelated diseases, but instead display focused association patterns within biologically and therapeutically coherent disease areas. This analysis challenges the notion that broad disease associations necessarily imply poor target specificity, highlighting instead that specificity often emerges at the level of therapeutic areas rather than individual diseases.
Figure 3. Disease class enrichment patterns for selected targets. Log2 enrichment scores show whether each target’s disease associations are concentrated within specific therapeutic areas. Positive values (pink) indicate enrichment, negative values (blue) indicate depletion compared to genome-wide expectations.
To what extent does DISGENET evidence support the approved clinical indications of these targets? Thirty-seven drugs show evidence supporting their approved indication, or a closely related disease phenotype (Figure 4).
Some targets are supported by strong and long-standing evidence. For example, PCSK9 displays maximal strength of association (DISGENET score 1) with hypercholesterolemia, fully consistent with its role as the therapeutic target of lerodalcibep (Lerochol). Similarly, F12 shows a strong association with hereditary angioedema, concordant with its targeting by garadacimab (Andembry), while PAH exhibits robust genetic support for hyperphenylalaninemia, fully aligned with its modulation by sepiapterin (Sephience). In these cases, molecular association data indicate that the targets are causally involved in the approved disease indication.
Other targets exhibit more moderate but biologically coherent support. CSF1R, targeted by vimseltinib (Romvimza) for tenosynovial giant cell tumors, is associated with fibrous histiocytic and related proliferative disorders, reflecting shared cellular origins and signaling pathways. Similarly, TRPM8, targeted by acoltremon (Tryptyr) for dry eye disease, shows molecular associations with dry eye syndromes and related sensory and epithelial dysfunctions. Collectively, these patterns suggest that regulatory success is often underpinned not only by direct molecular evidence for a specific indication, but also by a broader network of genetically and biologically related disease associations that support translational plausibility.
Figure 4. DISGENET evidence strength supporting approved drug indications. The heatmap shows DISGENET association scores between drug targets (rows) and diseases (columns). Color intensity represents the strength of molecular evidence, with gene symbols displayed in cells. White cells indicate no documented association.
Analysis of GTEx expression profiles revealed patterns in the tissue distribution of 2025 drug targets. A subset shows enriched expression in tissues directly relevant to their therapeutic indication. For example, PCSK9, APOC3, and PAH display strong liver-enriched expression, consistent with their roles in lipid metabolism and amino acid homeostasis. Similarly, MYBPC3 exhibits highly specific expression in cardiac muscle, aligning closely with its role in cardiomyopathy and supporting the paradigm that tissue-restricted expression can enhance on-target efficacy while limiting off-target effects.
A small subset of approved targets shows very low or near-absent expression across GTEx tissues. In most cases, this likely reflects limitations of bulk adult tissue transcriptomic data rather than true biological absence. Several targets are primarily expressed in rare cell populations or disease-activated states that are poorly captured in healthy reference atlases. For example, IL5 exhibits inducible expression in activated immune cells. In this case, disease context rather than baseline tissue expression underpins therapeutic relevance.
Conversely, other targets exhibit broad or near-ubiquitous expression across tissues. Overall, these findings indicate that tissue restriction is not a universal requirement for druggability. Instead, clinical success often depends on a balance between disease-relevant expression, pathway context, and the ability to achieve functional selectivity through dosing, delivery, or molecular mechanism.
Figure 5. Tissue expression profiles of targets for drugs approved in 2025. The heatmap shows median gene expression (log₁₀ transformed TPM) across GTEx tissues. Color intensity indicates relative expression levels within each gene, with hierarchical clustering applied to both tissues and genes.
To place the 2025 approvals in historical context, we examined prior drug and indication coverage for these targets using ChEMBL, excluding current-year approvals. The resulting landscape reveals heterogeneity in prior clinical maturity, ranging from targets with no previously approved drugs to highly exploited targets associated with dozens of compounds and indications. Well-established targets such as EGFR, ESR1, PDCD1, and CACNA1C sit at the highly saturated end of this spectrum, reflecting decades of iterative clinical exploration and extensive target reuse across disease areas.
Figure 6. Historical target maturity prior to 2025 FDA approvals. Distribution of drug and indication coverage for molecular targets derived from ChEMBL (excluding current-year approvals). Point size represents number of Phase IV drugs. Molecular support indicates whether there is information about the target-indication association in DISGENET.
In contrast, a subset of targets—including CLPP, CTSC, MASP2, and TNFSF13—shows little or no prior indication coverage despite substantial molecular and biological evidence, underscoring their relatively recent or niche clinical emergence.
Together, these patterns show that the 2025 approvals draw from a heterogeneous target landscape, where regulatory success arises both from long-standing clinical familiarity and from comparatively underexploited targets.
The 2025 FDA approvals illustrate that successful drug targets strike a balance between breadth and focus: they are neither narrowly confined to a single disease nor diffusely associated with unrelated conditions, but instead show broad molecular and disease associations anchored within a small number of biologically and therapeutically coherent disease classes. Most approved drugs are supported by molecular evidence that either directly matches the approved indication or aligns closely with related disease phenotypes, underscoring how strong target–indication concordance can reduce translational risk. Analyses based on DISGENET highlight how structured gene–disease association data can help distinguish true therapeutic coherence from superficial disease breadth, making it a valuable resource for interpreting target relevance in a systematic way.
While tissue-enriched expression often reinforces classical target-selection heuristics, it is not a universal requirement for druggability: many clinically successful targets are broadly expressed or appear weakly expressed in reference atlases, achieving selectivity instead through disease activation, cell-type restriction, or therapeutic modality.
Taken together, these findings emphasize that target selection is ultimately a systems-level decision, where integrating molecular evidence, disease-class specificity, expression context, and historical maturity provides a more robust framework than relying on any single metric alone.